Hostility, as measured by the Cook-Medley Hostility Scale of the Minnesota Multiphasic Personality Inventory, has been found to predict higher rates of both coronary heart disease and all-cause mortality. To evaluate one mechanism whereby hostility might contribute to health problems, the authors used regression models to determine whether hostility measured in college (1964-1966) predicted coronary risk factors assessed 21-23 years later (1987-1990) in 4,710 men and women. Of this group, 828 had lipids measured (1988-1991). Persons with higher hostility scores in college were significantly more likely at follow-up to consume more caffeine (r = 0.043), to have a larger body mass index (r = 0.055), to have higher lipid ratios (r = 0.092), and to be current smokers (r = 0.069) than those with lower hostility scores during college. Cross-sectional analyses found significant associations of contemporaneous hostility scores with the same four risk factors, as well as with alcohol consumption and hypertension (rs ranging from 0.043 to 0.117). These associations are large enough to have possible public health significance. We conclude that hostility may contribute to health problems through its influences on several coronary risk factors across the adult life span.